Artificial Intelligence & Blockchain — our perspective on potential synergies

Artificial Intelligence and blockchain are two of the most hyped technology topics all over the blogosphere. As our teams at Deutsche Telekom are dealing with the interlink of Data Management and Artificial Intelligence, we constantly think about how adjacent technologies like IoT or even Blockchain relate to our areas of expertise.

What we have recently observed is that Artificial Intelligence has truly arrived at the Board Level Agendas. Our confidence is high that the topic of #blockchain has certainly reached a similar hype level in almost all industries. At least from what we observe in Deutsche Telekom, we can clearly see that this is definitely the case.

Today, we would like to invite you to join our thinking about how Telecommunication companies such as Deutsche Telekom can benefit from synergies between Artificial Intelligence & blockchain. Thereby, we will first shortly summarize the key ideas behind blockchain (for experts, feel free to skip that). As a matter of completion, we reference to one of our earlier articles in which we excessively explained the key ideas behind Artificial intelligence. In a second step, we will mention how AI can benefit from blockchain as well as the other way around. More precisely, when talking about AI we focus specifically on machine learning and data availability/sharing.

Blockchain — a distributed ledger technology

A blockchain is a chronological list of elements, called “blocks”. These blocks are bound to each other. Each block contains a hash of the previous block, a timestamp and content information about the underlying matter. What benefits does blockchain bring? Let us say a group of people wanted to record transactions between each other. When using a blockchain approach, each peer would receive a copy of the blockchain, which is extended every time a new transaction happens. If one participant adds a block to the chain, other users immediately verify it. Thereby a single participant can hardly corrupt a blockchain, because the peers would detect it.

Sometimes, people tend to use the terms blockchain interchangeably with “Distributed Ledger Technology” (DLT). However, blockchain is (only) one special type of the DLT concept. Without getting too deep into the different approaches of DLT it is worth to mention differentiating properties such as whether being public or private, the used consensus algorithms (e.g. Proof of Work, Proof of Stake), or whether being mineable or not. Wecan highly recommend an article published by the World Bank, which gives a solid introduction into the details of DLT and Blockchain.

Deutsche Telekom actively drives Blockchain forward

The common benefit in all DLT approaches is that they disintermediate centralized administrators/brokers, because the peer group itself and not an intermediary ensures the validity of transactions. Thereby DLT has the potential to disrupt significant parts of value chains in many industries (such as banking) and possibly, it will become a commodity technology in the near future.

Since 2017 Deutsche Telekom Innovation Laboratories (T-Labs) has been exploring the possibilities of DLT and actively drives the topic forward. With T-Labs, Deutsche Telekom is researching and developing concepts with DLT, focusing on how ledger technology works by installing and experimenting on numerous ledger systems, such as Bitcoin, Ethereum, and IOTA. Additionally, T-Labs is formally joining a few foundations to push the technology forward, such as agreeing to become a steward of the Sovrin foundation (

Synergies of Artificial Intelligence (specifically Machine Learning) & Distributed Ledger Technology

As mentioned earlier, the key for artificially intelligent machines, enabled by Machine Learning, is the availability of and the efficient access to a wide range of high quality and reliable data. At the same time, decision makers that rely on the inference of machine learning models need to eventually trust and believe in the proclaimed values and directives. For both aspects, data acquisition as well as reliable machine learning models, DLT has the potential to add significant value.

In fact, DLT has the potential to tear down data silos. If DLT manages data access and ownership in the back, one could effectively solve a problem with machine learning. This is the case, because it would allow easier access to a higher volume and variety of data. Finally, this would eventually lead to better and more accurate model decisions. So if the wide introduction of DLT leads to an opening of silos and a clear and effective way to charge data access, companies like banks, insurances but also Telcos like Deutsche Telekom might be able to extend their internal scoring models with external data. This opens up the opportunity to lead to earlier identification of fraudulent or risky customers. Many other use cases are imaginable that benefit from the commoditization of data brokerage, which is enabled and secured by DLT.

How Data brokerage platforms may benefit from DLT

Especially for the data acquisition aspect, Deutsche Telekom currently develops new platforms and business models. One prominent external example is the so-called “Data Intelligence Hub” (DIH), that T-Systems has recently launched). As of today, in its first development stage the platform does not yet embody DLT. However, the use case and benefit of it should be straightforward. As DIH is a market place for the exchange of data, DLT would control so-called “smart contracts” that steer and record any data and model exchange between parties of the DIH. In this context, the ability of DLT to prevent identity theft and thereby ensure that trading parties can trust each other is an additional benefit DLT would bring to the table.

As data from different sources are shared and consolidated via a platform such as the Data Intelligence Hub, involved parties may apply machine learning techniques to train models that can itself be traded on the platform as well. Here again, DLT may mitigate these trades via smart contracts between the involved parties on the one hand. Furthermore, on the other hand, DLT can ensure transparency on what factors specifically influence the properties of a trained model. This would enable the ability to comprehend why a model infers certain values and decisions. In this case, one could think of a blockchain as a kind of “logchain”.

DLT as security leaver

To sum up, DLT can help to enable, track, understand and explain reliable decisions made by AI (specifically machine learning models). Additionally, in terms of security aspects, DLT has the potential to prevent machine learning models from getting skewed by fraudulent parties that forcefully add unbalanced data in order to influence a certain behavior of machines. Since such influence would be much easier to detect by other involved parties, no single entity could risk the humiliation. It becomes apparent that DLT has the potential to optimize data trading, as well as transparency on ownership of data and models. In the long term, DLT may ensure validity and trustworthiness of models and their inferred decisions.

The other way around

Now that we have shared our thoughts on how blockchain may fuel the establishment of data sharing and machine learning training, distribution and application, I want to mention shortly, that Machine Learning can help the DLT movement to achievement efficiency gains, too. To give a very basic example, everyone has heard about the mining process of bitcoins and that it takes more and more effort to mine these crypto-currencies. In fact, mining is in general an increasingly energy consuming process that is inherent in a lot DLT applications. Artificially intelligent machines have the potential to allocate resources to the mining process in a smarter way so that the overall energy consumption is lower. I want to recommend the paper of Marwala & Xing (2018) in which one can find an extensive list of how AI technologies can facilitate more secure and efficient DLT-based solutions.

And for more synergies between AI and Blockchain I highly recommend these two articles:

As always, please understand this blog post as an invitation to openly discuss thoughts on topics that are currently of high importance for me personally and the industry we work in. We look forward to engage in a discussion with you either personally or here on LinkedIn.

Written by Susan Wegner, with help from John Calian

Susan Wegner leads the Chief Data Office of Deutsche Telekom, and is based in Berlin.

John Calian leads the Deutsche Telekom Innovation Labs (T-Labs) in Berlin, and is also the lead on blockchain driven strategy, DT-wide.